diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,3930 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 100.0, + "eval_steps": 500, + "global_step": 10600, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 1.0, + "grad_norm": 1.6121073961257935, + "learning_rate": 4.9500000000000004e-05, + "loss": 1.0203, + "step": 106 + }, + { + "epoch": 1.0, + "eval_LOCATION_f1": 0.0, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.0, + "eval_LOCATION_recall": 0.0, + "eval_ORGANIZATION_f1": 0.0, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.0, + "eval_ORGANIZATION_recall": 0.0, + "eval_PERSON_f1": 0.03508771929824562, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.14285714285714285, + "eval_PERSON_recall": 0.02, + "eval_QUANTITY_f1": 0.0, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.0, + "eval_QUANTITY_recall": 0.0, + "eval_TIME_f1": 0.05405405405405406, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.3333333333333333, + "eval_TIME_recall": 0.029411764705882353, + "eval_loss": 0.591905415058136, + "eval_overall_accuracy": 0.840019689884322, + "eval_overall_f1": 0.020618556701030927, + "eval_overall_precision": 0.10810810810810811, + "eval_overall_recall": 0.011396011396011397, + "eval_runtime": 0.9779, + "eval_samples_per_second": 191.225, + "eval_steps_per_second": 3.068, + "step": 106 + }, + { + "epoch": 2.0, + "grad_norm": 1.089346170425415, + "learning_rate": 4.9e-05, + "loss": 0.5053, + "step": 212 + }, + { + "epoch": 2.0, + "eval_LOCATION_f1": 0.2777777777777778, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.42857142857142855, + "eval_LOCATION_recall": 0.2054794520547945, + "eval_ORGANIZATION_f1": 0.06557377049180328, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.07017543859649122, + "eval_ORGANIZATION_recall": 0.06153846153846154, + "eval_PERSON_f1": 0.6512968299711815, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.5736040609137056, + "eval_PERSON_recall": 0.7533333333333333, + "eval_QUANTITY_f1": 0.08888888888888889, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.125, + "eval_QUANTITY_recall": 0.06896551724137931, + "eval_TIME_f1": 0.5074626865671642, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.5151515151515151, + "eval_TIME_recall": 0.5, + "eval_loss": 0.3727489709854126, + "eval_overall_accuracy": 0.8951513659857249, + "eval_overall_f1": 0.4383164005805515, + "eval_overall_precision": 0.4467455621301775, + "eval_overall_recall": 0.4301994301994302, + "eval_runtime": 1.0224, + "eval_samples_per_second": 182.9, + "eval_steps_per_second": 2.934, + "step": 212 + }, + { + "epoch": 3.0, + "grad_norm": 0.9725346565246582, + "learning_rate": 4.85e-05, + "loss": 0.3221, + "step": 318 + }, + { + "epoch": 3.0, + "eval_LOCATION_f1": 0.6227544910179641, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.5531914893617021, + "eval_LOCATION_recall": 0.7123287671232876, + "eval_ORGANIZATION_f1": 0.2689075630252101, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.2962962962962963, + "eval_ORGANIZATION_recall": 0.24615384615384617, + "eval_PERSON_f1": 0.8145896656534953, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.7486033519553073, + "eval_PERSON_recall": 0.8933333333333333, + "eval_QUANTITY_f1": 0.5588235294117647, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.48717948717948717, + "eval_QUANTITY_recall": 0.6551724137931034, + "eval_TIME_f1": 0.8169014084507041, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7837837837837838, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.21981501579284668, + "eval_overall_accuracy": 0.9305931577651981, + "eval_overall_f1": 0.6631299734748011, + "eval_overall_precision": 0.6203473945409429, + "eval_overall_recall": 0.7122507122507122, + "eval_runtime": 1.0345, + "eval_samples_per_second": 180.757, + "eval_steps_per_second": 2.9, + "step": 318 + }, + { + "epoch": 4.0, + "grad_norm": 0.8557797074317932, + "learning_rate": 4.8e-05, + "loss": 0.2222, + "step": 424 + }, + { + "epoch": 4.0, + "eval_LOCATION_f1": 0.6709677419354839, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.6341463414634146, + "eval_LOCATION_recall": 0.7123287671232876, + "eval_ORGANIZATION_f1": 0.5844155844155845, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.5056179775280899, + "eval_ORGANIZATION_recall": 0.6923076923076923, + "eval_PERSON_f1": 0.8411214953271027, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.7894736842105263, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.6567164179104478, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.5789473684210527, + "eval_QUANTITY_recall": 0.7586206896551724, + "eval_TIME_f1": 0.8219178082191781, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7692307692307693, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.17654158174991608, + "eval_overall_accuracy": 0.9416687176962836, + "eval_overall_f1": 0.7376623376623377, + "eval_overall_precision": 0.6778042959427207, + "eval_overall_recall": 0.8091168091168092, + "eval_runtime": 0.9921, + "eval_samples_per_second": 188.484, + "eval_steps_per_second": 3.024, + "step": 424 + }, + { + "epoch": 5.0, + "grad_norm": 1.1405649185180664, + "learning_rate": 4.75e-05, + "loss": 0.1869, + "step": 530 + }, + { + "epoch": 5.0, + "eval_LOCATION_f1": 0.7682926829268293, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.6923076923076923, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.6296296296296297, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.5257731958762887, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8490566037735849, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8035714285714286, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.6086956521739131, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.525, + "eval_QUANTITY_recall": 0.7241379310344828, + "eval_TIME_f1": 0.7792207792207793, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6976744186046512, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.15994760394096375, + "eval_overall_accuracy": 0.9473295594388383, + "eval_overall_f1": 0.759493670886076, + "eval_overall_precision": 0.683371298405467, + "eval_overall_recall": 0.8547008547008547, + "eval_runtime": 0.9628, + "eval_samples_per_second": 194.219, + "eval_steps_per_second": 3.116, + "step": 530 + }, + { + "epoch": 6.0, + "grad_norm": 1.9796879291534424, + "learning_rate": 4.7e-05, + "loss": 0.169, + "step": 636 + }, + { + "epoch": 6.0, + "eval_LOCATION_f1": 0.6993865030674846, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.6333333333333333, + "eval_LOCATION_recall": 0.7808219178082192, + "eval_ORGANIZATION_f1": 0.6153846153846154, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.5, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.8769716088328076, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8323353293413174, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.5897435897435898, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.46938775510204084, + "eval_QUANTITY_recall": 0.7931034482758621, + "eval_TIME_f1": 0.7073170731707317, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6041666666666666, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.19615504145622253, + "eval_overall_accuracy": 0.9318237755353187, + "eval_overall_f1": 0.7416563658838071, + "eval_overall_precision": 0.6550218340611353, + "eval_overall_recall": 0.8547008547008547, + "eval_runtime": 0.9795, + "eval_samples_per_second": 190.923, + "eval_steps_per_second": 3.063, + "step": 636 + }, + { + "epoch": 7.0, + "grad_norm": 0.9608461260795593, + "learning_rate": 4.6500000000000005e-05, + "loss": 0.154, + "step": 742 + }, + { + "epoch": 7.0, + "eval_LOCATION_f1": 0.7560975609756097, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.6813186813186813, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.6620689655172414, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6, + "eval_ORGANIZATION_recall": 0.7384615384615385, + "eval_PERSON_f1": 0.8817891373801918, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8466257668711656, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.6956521739130435, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7532467532467532, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6744186046511628, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.14681145548820496, + "eval_overall_accuracy": 0.9492985478710313, + "eval_overall_f1": 0.7838541666666667, + "eval_overall_precision": 0.7218225419664268, + "eval_overall_recall": 0.8575498575498576, + "eval_runtime": 0.9965, + "eval_samples_per_second": 187.649, + "eval_steps_per_second": 3.01, + "step": 742 + }, + { + "epoch": 8.0, + "grad_norm": 0.8899044394493103, + "learning_rate": 4.600000000000001e-05, + "loss": 0.1439, + "step": 848 + }, + { + "epoch": 8.0, + "eval_LOCATION_f1": 0.7354838709677419, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.6951219512195121, + "eval_LOCATION_recall": 0.7808219178082192, + "eval_ORGANIZATION_f1": 0.6887417218543046, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6046511627906976, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.879746835443038, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8373493975903614, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.716417910447761, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.631578947368421, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7733333333333334, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7073170731707317, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1457252949476242, + "eval_overall_accuracy": 0.9505291656411519, + "eval_overall_f1": 0.787958115183246, + "eval_overall_precision": 0.7288135593220338, + "eval_overall_recall": 0.8575498575498576, + "eval_runtime": 1.0009, + "eval_samples_per_second": 186.838, + "eval_steps_per_second": 2.997, + "step": 848 + }, + { + "epoch": 9.0, + "grad_norm": 1.0214319229125977, + "learning_rate": 4.55e-05, + "loss": 0.1317, + "step": 954 + }, + { + "epoch": 9.0, + "eval_LOCATION_f1": 0.7870967741935484, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7439024390243902, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.6585365853658536, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.5454545454545454, + "eval_ORGANIZATION_recall": 0.8307692307692308, + "eval_PERSON_f1": 0.887459807073955, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8571428571428571, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7272727272727273, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6486486486486487, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7837837837837837, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.725, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1383334845304489, + "eval_overall_accuracy": 0.9524981540733448, + "eval_overall_f1": 0.7948051948051948, + "eval_overall_precision": 0.7303102625298329, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 1.0066, + "eval_samples_per_second": 185.771, + "eval_steps_per_second": 2.98, + "step": 954 + }, + { + "epoch": 10.0, + "grad_norm": 0.8760809302330017, + "learning_rate": 4.5e-05, + "loss": 0.1261, + "step": 1060 + }, + { + "epoch": 10.0, + "eval_LOCATION_f1": 0.8235294117647058, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7875, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.6887417218543046, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6046511627906976, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.8810289389067525, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8509316770186336, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7076923076923076, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6388888888888888, + "eval_QUANTITY_recall": 0.7931034482758621, + "eval_TIME_f1": 0.8219178082191781, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7692307692307693, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.1302676945924759, + "eval_overall_accuracy": 0.9559438838296825, + "eval_overall_f1": 0.8100929614873837, + "eval_overall_precision": 0.7587064676616916, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 0.9994, + "eval_samples_per_second": 187.11, + "eval_steps_per_second": 3.002, + "step": 1060 + }, + { + "epoch": 11.0, + "grad_norm": 0.8754971027374268, + "learning_rate": 4.4500000000000004e-05, + "loss": 0.1191, + "step": 1166 + }, + { + "epoch": 11.0, + "eval_LOCATION_f1": 0.781456953642384, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7564102564102564, + "eval_LOCATION_recall": 0.8082191780821918, + "eval_ORGANIZATION_f1": 0.7234042553191491, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6710526315789473, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.887459807073955, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8571428571428571, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.676923076923077, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6111111111111112, + "eval_QUANTITY_recall": 0.7586206896551724, + "eval_TIME_f1": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.12534178793430328, + "eval_overall_accuracy": 0.9596357371400444, + "eval_overall_f1": 0.8070175438596491, + "eval_overall_precision": 0.7666666666666667, + "eval_overall_recall": 0.8518518518518519, + "eval_runtime": 0.9868, + "eval_samples_per_second": 189.5, + "eval_steps_per_second": 3.04, + "step": 1166 + }, + { + "epoch": 12.0, + "grad_norm": 0.8746627569198608, + "learning_rate": 4.4000000000000006e-05, + "loss": 0.1131, + "step": 1272 + }, + { + "epoch": 12.0, + "eval_LOCATION_f1": 0.7361963190184049, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.6666666666666666, + "eval_LOCATION_recall": 0.821917808219178, + "eval_ORGANIZATION_f1": 0.7051282051282052, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6043956043956044, + "eval_ORGANIZATION_recall": 0.8461538461538461, + "eval_PERSON_f1": 0.8910256410256411, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8580246913580247, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.7575757575757576, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6756756756756757, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.7532467532467532, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6744186046511628, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1429539918899536, + "eval_overall_accuracy": 0.9537287718434654, + "eval_overall_f1": 0.7958656330749354, + "eval_overall_precision": 0.7281323877068558, + "eval_overall_recall": 0.8774928774928775, + "eval_runtime": 1.0275, + "eval_samples_per_second": 181.996, + "eval_steps_per_second": 2.92, + "step": 1272 + }, + { + "epoch": 13.0, + "grad_norm": 0.7008482813835144, + "learning_rate": 4.35e-05, + "loss": 0.1101, + "step": 1378 + }, + { + "epoch": 13.0, + "eval_LOCATION_f1": 0.7999999999999999, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7792207792207793, + "eval_LOCATION_recall": 0.821917808219178, + "eval_ORGANIZATION_f1": 0.728476821192053, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6395348837209303, + "eval_ORGANIZATION_recall": 0.8461538461538461, + "eval_PERSON_f1": 0.8938906752411575, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8633540372670807, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.6376811594202898, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.55, + "eval_QUANTITY_recall": 0.7586206896551724, + "eval_TIME_f1": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.13461139798164368, + "eval_overall_accuracy": 0.9564361309377307, + "eval_overall_f1": 0.8090185676392573, + "eval_overall_precision": 0.7568238213399504, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 0.9921, + "eval_samples_per_second": 188.484, + "eval_steps_per_second": 3.024, + "step": 1378 + }, + { + "epoch": 14.0, + "grad_norm": 0.6436917185783386, + "learning_rate": 4.3e-05, + "loss": 0.105, + "step": 1484 + }, + { + "epoch": 14.0, + "eval_LOCATION_f1": 0.7848101265822783, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7294117647058823, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7234042553191491, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6710526315789473, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8932038834951457, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8679245283018868, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.746268656716418, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6578947368421053, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.7837837837837837, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.725, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.13453252613544464, + "eval_overall_accuracy": 0.9591434900319961, + "eval_overall_f1": 0.814419225634179, + "eval_overall_precision": 0.7663316582914573, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 0.9965, + "eval_samples_per_second": 187.656, + "eval_steps_per_second": 3.011, + "step": 1484 + }, + { + "epoch": 15.0, + "grad_norm": 1.208773136138916, + "learning_rate": 4.25e-05, + "loss": 0.1043, + "step": 1590 + }, + { + "epoch": 15.0, + "eval_LOCATION_f1": 0.7843137254901962, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.75, + "eval_LOCATION_recall": 0.821917808219178, + "eval_ORGANIZATION_f1": 0.7412587412587412, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6794871794871795, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.8990228013029316, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8789808917197452, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7575757575757576, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6756756756756757, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1265690177679062, + "eval_overall_accuracy": 0.961112478464189, + "eval_overall_f1": 0.8221024258760108, + "eval_overall_precision": 0.7800511508951407, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 1.0073, + "eval_samples_per_second": 185.651, + "eval_steps_per_second": 2.978, + "step": 1590 + }, + { + "epoch": 16.0, + "grad_norm": 1.1977620124816895, + "learning_rate": 4.2e-05, + "loss": 0.0979, + "step": 1696 + }, + { + "epoch": 16.0, + "eval_LOCATION_f1": 0.7792207792207793, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7407407407407407, + "eval_LOCATION_recall": 0.821917808219178, + "eval_ORGANIZATION_f1": 0.7236842105263158, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.632183908045977, + "eval_ORGANIZATION_recall": 0.8461538461538461, + "eval_PERSON_f1": 0.90032154340836, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8695652173913043, + "eval_PERSON_recall": 0.9333333333333333, + "eval_QUANTITY_f1": 0.6666666666666666, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.5581395348837209, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7368421052631577, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6666666666666666, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.1487545669078827, + "eval_overall_accuracy": 0.9527442776273689, + "eval_overall_f1": 0.8026143790849672, + "eval_overall_precision": 0.7415458937198067, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.9958, + "eval_samples_per_second": 187.787, + "eval_steps_per_second": 3.013, + "step": 1696 + }, + { + "epoch": 17.0, + "grad_norm": 0.41393932700157166, + "learning_rate": 4.15e-05, + "loss": 0.0954, + "step": 1802 + }, + { + "epoch": 17.0, + "eval_LOCATION_f1": 0.8104575163398693, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.775, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7027027027027027, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6265060240963856, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.90032154340836, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8695652173913043, + "eval_PERSON_recall": 0.9333333333333333, + "eval_QUANTITY_f1": 0.7619047619047619, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7058823529411765, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7631578947368423, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.6904761904761905, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.13488559424877167, + "eval_overall_accuracy": 0.958897366477972, + "eval_overall_f1": 0.8175765645805593, + "eval_overall_precision": 0.7675, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.9952, + "eval_samples_per_second": 187.902, + "eval_steps_per_second": 3.014, + "step": 1802 + }, + { + "epoch": 18.0, + "grad_norm": 1.0734295845031738, + "learning_rate": 4.1e-05, + "loss": 0.0928, + "step": 1908 + }, + { + "epoch": 18.0, + "eval_LOCATION_f1": 0.7870967741935484, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7439024390243902, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.726027397260274, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.654320987654321, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.8838709677419355, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.85625, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7692307692307692, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6944444444444444, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.7837837837837837, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.725, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1342119574546814, + "eval_overall_accuracy": 0.9596357371400444, + "eval_overall_f1": 0.8133333333333332, + "eval_overall_precision": 0.7644110275689223, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 1.0888, + "eval_samples_per_second": 171.747, + "eval_steps_per_second": 2.755, + "step": 1908 + }, + { + "epoch": 19.0, + "grad_norm": 0.5558896660804749, + "learning_rate": 4.05e-05, + "loss": 0.0884, + "step": 2014 + }, + { + "epoch": 19.0, + "eval_LOCATION_f1": 0.8157894736842105, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7848101265822784, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7042253521126761, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6493506493506493, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.75, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6857142857142857, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.12890850007534027, + "eval_overall_accuracy": 0.9616047255722373, + "eval_overall_f1": 0.8200270635994589, + "eval_overall_precision": 0.7809278350515464, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 0.9958, + "eval_samples_per_second": 187.787, + "eval_steps_per_second": 3.013, + "step": 2014 + }, + { + "epoch": 20.0, + "grad_norm": 0.4977124333381653, + "learning_rate": 4e-05, + "loss": 0.0851, + "step": 2120 + }, + { + "epoch": 20.0, + "eval_LOCATION_f1": 0.7692307692307693, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7228915662650602, + "eval_LOCATION_recall": 0.821917808219178, + "eval_ORGANIZATION_f1": 0.7412587412587412, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6794871794871795, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.8903225806451613, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8625, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7619047619047619, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7058823529411765, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7837837837837837, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.725, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.137658029794693, + "eval_overall_accuracy": 0.9616047255722373, + "eval_overall_f1": 0.8150134048257373, + "eval_overall_precision": 0.769620253164557, + "eval_overall_recall": 0.8660968660968661, + "eval_runtime": 0.9955, + "eval_samples_per_second": 187.845, + "eval_steps_per_second": 3.014, + "step": 2120 + }, + { + "epoch": 21.0, + "grad_norm": 2.0087778568267822, + "learning_rate": 3.9500000000000005e-05, + "loss": 0.0819, + "step": 2226 + }, + { + "epoch": 21.0, + "eval_LOCATION_f1": 0.8235294117647058, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7875, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7361111111111112, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6708860759493671, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7619047619047619, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7058823529411765, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.12961311638355255, + "eval_overall_accuracy": 0.9633275904504061, + "eval_overall_f1": 0.8319783197831978, + "eval_overall_precision": 0.7932816537467701, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 1.0079, + "eval_samples_per_second": 185.535, + "eval_steps_per_second": 2.976, + "step": 2226 + }, + { + "epoch": 22.0, + "grad_norm": 1.2587146759033203, + "learning_rate": 3.9000000000000006e-05, + "loss": 0.0858, + "step": 2332 + }, + { + "epoch": 22.0, + "eval_LOCATION_f1": 0.7999999999999999, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7792207792207793, + "eval_LOCATION_recall": 0.821917808219178, + "eval_ORGANIZATION_f1": 0.7199999999999999, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6352941176470588, + "eval_ORGANIZATION_recall": 0.8307692307692308, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.8064516129032258, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7575757575757576, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.13012367486953735, + "eval_overall_accuracy": 0.9613586020182131, + "eval_overall_f1": 0.8281461434370772, + "eval_overall_precision": 0.788659793814433, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.9889, + "eval_samples_per_second": 189.099, + "eval_steps_per_second": 3.034, + "step": 2332 + }, + { + "epoch": 23.0, + "grad_norm": 1.0316872596740723, + "learning_rate": 3.85e-05, + "loss": 0.0789, + "step": 2438 + }, + { + "epoch": 23.0, + "eval_LOCATION_f1": 0.7922077922077922, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7530864197530864, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7391304347826088, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6986301369863014, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.9061488673139159, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8805031446540881, + "eval_PERSON_recall": 0.9333333333333333, + "eval_QUANTITY_f1": 0.7936507936507937, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7352941176470589, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8169014084507041, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7837837837837838, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.13557566702365875, + "eval_overall_accuracy": 0.9628353433423579, + "eval_overall_f1": 0.8326530612244898, + "eval_overall_precision": 0.796875, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.9917, + "eval_samples_per_second": 188.567, + "eval_steps_per_second": 3.025, + "step": 2438 + }, + { + "epoch": 24.0, + "grad_norm": 0.8608385324478149, + "learning_rate": 3.8e-05, + "loss": 0.0763, + "step": 2544 + }, + { + "epoch": 24.0, + "eval_LOCATION_f1": 0.7894736842105263, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.759493670886076, + "eval_LOCATION_recall": 0.821917808219178, + "eval_ORGANIZATION_f1": 0.7162162162162162, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6385542168674698, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.9025974025974027, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.879746835443038, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.8064516129032258, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7575757575757576, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1364046037197113, + "eval_overall_accuracy": 0.9613586020182131, + "eval_overall_f1": 0.8236877523553163, + "eval_overall_precision": 0.7806122448979592, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.9754, + "eval_samples_per_second": 191.709, + "eval_steps_per_second": 3.076, + "step": 2544 + }, + { + "epoch": 25.0, + "grad_norm": 0.9957919716835022, + "learning_rate": 3.7500000000000003e-05, + "loss": 0.0734, + "step": 2650 + }, + { + "epoch": 25.0, + "eval_LOCATION_f1": 0.7712418300653595, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7375, + "eval_LOCATION_recall": 0.8082191780821918, + "eval_ORGANIZATION_f1": 0.7375886524822695, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6842105263157895, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.9032258064516129, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.875, + "eval_PERSON_recall": 0.9333333333333333, + "eval_QUANTITY_f1": 0.75, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6857142857142857, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7733333333333334, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7073170731707317, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1437826156616211, + "eval_overall_accuracy": 0.961112478464189, + "eval_overall_f1": 0.8183041722745626, + "eval_overall_precision": 0.7755102040816326, + "eval_overall_recall": 0.8660968660968661, + "eval_runtime": 0.9957, + "eval_samples_per_second": 187.814, + "eval_steps_per_second": 3.013, + "step": 2650 + }, + { + "epoch": 26.0, + "grad_norm": 0.6454270482063293, + "learning_rate": 3.7e-05, + "loss": 0.0731, + "step": 2756 + }, + { + "epoch": 26.0, + "eval_LOCATION_f1": 0.7712418300653595, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7375, + "eval_LOCATION_recall": 0.8082191780821918, + "eval_ORGANIZATION_f1": 0.7092198581560285, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6578947368421053, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.896774193548387, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.86875, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.7272727272727273, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6486486486486487, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.14986544847488403, + "eval_overall_accuracy": 0.958897366477972, + "eval_overall_f1": 0.8102288021534321, + "eval_overall_precision": 0.7678571428571429, + "eval_overall_recall": 0.8575498575498576, + "eval_runtime": 0.9887, + "eval_samples_per_second": 189.131, + "eval_steps_per_second": 3.034, + "step": 2756 + }, + { + "epoch": 27.0, + "grad_norm": 0.7397086024284363, + "learning_rate": 3.65e-05, + "loss": 0.0704, + "step": 2862 + }, + { + "epoch": 27.0, + "eval_LOCATION_f1": 0.7973856209150326, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7625, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7375886524822695, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6842105263157895, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.8853503184713376, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8475609756097561, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.716417910447761, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.631578947368421, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.767123287671233, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.717948717948718, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.1502029001712799, + "eval_overall_accuracy": 0.9591434900319961, + "eval_overall_f1": 0.8128342245989305, + "eval_overall_precision": 0.7657430730478589, + "eval_overall_recall": 0.8660968660968661, + "eval_runtime": 0.9947, + "eval_samples_per_second": 188.003, + "eval_steps_per_second": 3.016, + "step": 2862 + }, + { + "epoch": 28.0, + "grad_norm": 1.38124418258667, + "learning_rate": 3.6e-05, + "loss": 0.0679, + "step": 2968 + }, + { + "epoch": 28.0, + "eval_LOCATION_f1": 0.794701986754967, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7692307692307693, + "eval_LOCATION_recall": 0.821917808219178, + "eval_ORGANIZATION_f1": 0.75, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6835443037974683, + "eval_ORGANIZATION_recall": 0.8307692307692308, + "eval_PERSON_f1": 0.8996763754045306, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8742138364779874, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.71875, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6571428571428571, + "eval_QUANTITY_recall": 0.7931034482758621, + "eval_TIME_f1": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.14466358721256256, + "eval_overall_accuracy": 0.9618508491262614, + "eval_overall_f1": 0.8232118758434548, + "eval_overall_precision": 0.782051282051282, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 0.9878, + "eval_samples_per_second": 189.3, + "eval_steps_per_second": 3.037, + "step": 2968 + }, + { + "epoch": 29.0, + "grad_norm": 0.8041614294052124, + "learning_rate": 3.55e-05, + "loss": 0.0657, + "step": 3074 + }, + { + "epoch": 29.0, + "eval_LOCATION_f1": 0.8211920529801324, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7948717948717948, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7346938775510204, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6585365853658537, + "eval_ORGANIZATION_recall": 0.8307692307692308, + "eval_PERSON_f1": 0.894736842105263, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8831168831168831, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7619047619047619, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7058823529411765, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.9117647058823528, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.9117647058823529, + "eval_TIME_recall": 0.9117647058823529, + "eval_loss": 0.1336374580860138, + "eval_overall_accuracy": 0.9645582082205267, + "eval_overall_f1": 0.8376534788540245, + "eval_overall_precision": 0.8036649214659686, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 0.9977, + "eval_samples_per_second": 187.429, + "eval_steps_per_second": 3.007, + "step": 3074 + }, + { + "epoch": 30.0, + "grad_norm": 0.9247944951057434, + "learning_rate": 3.5e-05, + "loss": 0.067, + "step": 3180 + }, + { + "epoch": 30.0, + "eval_LOCATION_f1": 0.7919463087248322, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7763157894736842, + "eval_LOCATION_recall": 0.8082191780821918, + "eval_ORGANIZATION_f1": 0.7210884353741497, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6463414634146342, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.8990228013029316, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8789808917197452, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.75, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6857142857142857, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.14372512698173523, + "eval_overall_accuracy": 0.9613586020182131, + "eval_overall_f1": 0.8222523744911805, + "eval_overall_precision": 0.7849740932642487, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 1.0038, + "eval_samples_per_second": 186.284, + "eval_steps_per_second": 2.989, + "step": 3180 + }, + { + "epoch": 31.0, + "grad_norm": 0.41009989380836487, + "learning_rate": 3.45e-05, + "loss": 0.0649, + "step": 3286 + }, + { + "epoch": 31.0, + "eval_LOCATION_f1": 0.8205128205128205, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7710843373493976, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7352941176470589, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.704225352112676, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.9019607843137256, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8846153846153846, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.75, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6857142857142857, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7777777777777778, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7368421052631579, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.1421852558851242, + "eval_overall_accuracy": 0.9625892197883338, + "eval_overall_f1": 0.8283378746594006, + "eval_overall_precision": 0.793733681462141, + "eval_overall_recall": 0.8660968660968661, + "eval_runtime": 0.9943, + "eval_samples_per_second": 188.063, + "eval_steps_per_second": 3.017, + "step": 3286 + }, + { + "epoch": 32.0, + "grad_norm": 0.4342913031578064, + "learning_rate": 3.4000000000000007e-05, + "loss": 0.0619, + "step": 3392 + }, + { + "epoch": 32.0, + "eval_LOCATION_f1": 0.7792207792207793, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7407407407407407, + "eval_LOCATION_recall": 0.821917808219178, + "eval_ORGANIZATION_f1": 0.7236842105263158, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.632183908045977, + "eval_ORGANIZATION_recall": 0.8461538461538461, + "eval_PERSON_f1": 0.896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8734177215189873, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.75, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6857142857142857, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8571428571428571, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8333333333333334, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.1523984968662262, + "eval_overall_accuracy": 0.9593896135860202, + "eval_overall_f1": 0.8208556149732621, + "eval_overall_precision": 0.7732997481108312, + "eval_overall_recall": 0.8746438746438746, + "eval_runtime": 1.0018, + "eval_samples_per_second": 186.663, + "eval_steps_per_second": 2.995, + "step": 3392 + }, + { + "epoch": 33.0, + "grad_norm": 0.8614174127578735, + "learning_rate": 3.35e-05, + "loss": 0.0616, + "step": 3498 + }, + { + "epoch": 33.0, + "eval_LOCATION_f1": 0.8133333333333332, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7922077922077922, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.72992700729927, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6944444444444444, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8925081433224756, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8726114649681529, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.75, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6857142857142857, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8333333333333333, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7894736842105263, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.14742247760295868, + "eval_overall_accuracy": 0.9623430962343096, + "eval_overall_f1": 0.8273972602739725, + "eval_overall_precision": 0.7968337730870713, + "eval_overall_recall": 0.8603988603988604, + "eval_runtime": 0.9963, + "eval_samples_per_second": 187.698, + "eval_steps_per_second": 3.011, + "step": 3498 + }, + { + "epoch": 34.0, + "grad_norm": 1.166977882385254, + "learning_rate": 3.3e-05, + "loss": 0.0597, + "step": 3604 + }, + { + "epoch": 34.0, + "eval_LOCATION_f1": 0.8181818181818182, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7777777777777778, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7333333333333334, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6470588235294118, + "eval_ORGANIZATION_recall": 0.8461538461538461, + "eval_PERSON_f1": 0.8838709677419355, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.85625, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7936507936507937, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7352941176470589, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8055555555555555, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7631578947368421, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.14954759180545807, + "eval_overall_accuracy": 0.958897366477972, + "eval_overall_f1": 0.8251001335113485, + "eval_overall_precision": 0.7763819095477387, + "eval_overall_recall": 0.8803418803418803, + "eval_runtime": 1.0032, + "eval_samples_per_second": 186.407, + "eval_steps_per_second": 2.99, + "step": 3604 + }, + { + "epoch": 35.0, + "grad_norm": 0.7428762912750244, + "learning_rate": 3.2500000000000004e-05, + "loss": 0.0594, + "step": 3710 + }, + { + "epoch": 35.0, + "eval_LOCATION_f1": 0.7820512820512822, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7349397590361446, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7361111111111112, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6708860759493671, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.8753993610223643, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8404907975460123, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.75, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6857142857142857, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7945205479452054, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7435897435897436, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.15552844107151031, + "eval_overall_accuracy": 0.9581589958158996, + "eval_overall_f1": 0.8106666666666666, + "eval_overall_precision": 0.7619047619047619, + "eval_overall_recall": 0.8660968660968661, + "eval_runtime": 0.9939, + "eval_samples_per_second": 188.147, + "eval_steps_per_second": 3.018, + "step": 3710 + }, + { + "epoch": 36.0, + "grad_norm": 0.8952577710151672, + "learning_rate": 3.2000000000000005e-05, + "loss": 0.0572, + "step": 3816 + }, + { + "epoch": 36.0, + "eval_LOCATION_f1": 0.8157894736842105, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7848101265822784, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7272727272727272, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6666666666666666, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.8852459016393444, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8709677419354839, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8450704225352113, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8108108108108109, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.15078987181186676, + "eval_overall_accuracy": 0.9625892197883338, + "eval_overall_f1": 0.8267394270122783, + "eval_overall_precision": 0.7931937172774869, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 1.0019, + "eval_samples_per_second": 186.652, + "eval_steps_per_second": 2.994, + "step": 3816 + }, + { + "epoch": 37.0, + "grad_norm": 1.4861292839050293, + "learning_rate": 3.15e-05, + "loss": 0.0543, + "step": 3922 + }, + { + "epoch": 37.0, + "eval_LOCATION_f1": 0.8181818181818182, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7777777777777778, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.6625766871165644, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.5510204081632653, + "eval_ORGANIZATION_recall": 0.8307692307692308, + "eval_PERSON_f1": 0.8802588996763754, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8553459119496856, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.75, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6857142857142857, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8169014084507041, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7837837837837838, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.16852639615535736, + "eval_overall_accuracy": 0.9549593896135861, + "eval_overall_f1": 0.8042049934296978, + "eval_overall_precision": 0.7463414634146341, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.9916, + "eval_samples_per_second": 188.593, + "eval_steps_per_second": 3.026, + "step": 3922 + }, + { + "epoch": 38.0, + "grad_norm": 0.8263400197029114, + "learning_rate": 3.1e-05, + "loss": 0.055, + "step": 4028 + }, + { + "epoch": 38.0, + "eval_LOCATION_f1": 0.794701986754967, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7692307692307693, + "eval_LOCATION_recall": 0.821917808219178, + "eval_ORGANIZATION_f1": 0.7183098591549296, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6623376623376623, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8859934853420196, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8662420382165605, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7619047619047619, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7058823529411765, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.861111111111111, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8157894736842105, + "eval_TIME_recall": 0.9117647058823529, + "eval_loss": 0.15474605560302734, + "eval_overall_accuracy": 0.9616047255722373, + "eval_overall_f1": 0.8217687074829932, + "eval_overall_precision": 0.7864583333333334, + "eval_overall_recall": 0.8603988603988604, + "eval_runtime": 0.9962, + "eval_samples_per_second": 187.719, + "eval_steps_per_second": 3.012, + "step": 4028 + }, + { + "epoch": 39.0, + "grad_norm": 0.7035608291625977, + "learning_rate": 3.05e-05, + "loss": 0.051, + "step": 4134 + }, + { + "epoch": 39.0, + "eval_LOCATION_f1": 0.832214765100671, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8157894736842105, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7517730496453899, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6973684210526315, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.8838709677419355, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.85625, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7868852459016394, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.75, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8450704225352113, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8108108108108109, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.14542774856090546, + "eval_overall_accuracy": 0.9645582082205267, + "eval_overall_f1": 0.8360655737704918, + "eval_overall_precision": 0.8031496062992126, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.9872, + "eval_samples_per_second": 189.427, + "eval_steps_per_second": 3.039, + "step": 4134 + }, + { + "epoch": 40.0, + "grad_norm": 0.368028461933136, + "learning_rate": 3e-05, + "loss": 0.053, + "step": 4240 + }, + { + "epoch": 40.0, + "eval_LOCATION_f1": 0.8133333333333332, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7922077922077922, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7066666666666667, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6235294117647059, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.8823529411764707, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.7619047619047619, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7058823529411765, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.7999999999999999, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7777777777777778, + "eval_TIME_recall": 0.8235294117647058, + "eval_loss": 0.15164527297019958, + "eval_overall_accuracy": 0.9598818606940684, + "eval_overall_f1": 0.8146143437077131, + "eval_overall_precision": 0.7757731958762887, + "eval_overall_recall": 0.8575498575498576, + "eval_runtime": 1.0735, + "eval_samples_per_second": 174.198, + "eval_steps_per_second": 2.795, + "step": 4240 + }, + { + "epoch": 41.0, + "grad_norm": 0.21483705937862396, + "learning_rate": 2.95e-05, + "loss": 0.0526, + "step": 4346 + }, + { + "epoch": 41.0, + "eval_LOCATION_f1": 0.7792207792207793, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7407407407407407, + "eval_LOCATION_recall": 0.821917808219178, + "eval_ORGANIZATION_f1": 0.7183098591549296, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6623376623376623, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8794788273615636, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8598726114649682, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.75, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6857142857142857, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8055555555555555, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7631578947368421, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.14916305243968964, + "eval_overall_accuracy": 0.9591434900319961, + "eval_overall_f1": 0.8092016238159675, + "eval_overall_precision": 0.770618556701031, + "eval_overall_recall": 0.8518518518518519, + "eval_runtime": 0.9975, + "eval_samples_per_second": 187.465, + "eval_steps_per_second": 3.007, + "step": 4346 + }, + { + "epoch": 42.0, + "grad_norm": 0.46089088916778564, + "learning_rate": 2.9e-05, + "loss": 0.0497, + "step": 4452 + }, + { + "epoch": 42.0, + "eval_LOCATION_f1": 0.8299319727891157, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8243243243243243, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7346938775510204, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6585365853658537, + "eval_ORGANIZATION_recall": 0.8307692307692308, + "eval_PERSON_f1": 0.8766233766233766, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8544303797468354, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.7936507936507937, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7352941176470589, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8169014084507041, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7837837837837838, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.14921055734157562, + "eval_overall_accuracy": 0.963081466896382, + "eval_overall_f1": 0.8260869565217391, + "eval_overall_precision": 0.7896103896103897, + "eval_overall_recall": 0.8660968660968661, + "eval_runtime": 0.9983, + "eval_samples_per_second": 187.314, + "eval_steps_per_second": 3.005, + "step": 4452 + }, + { + "epoch": 43.0, + "grad_norm": 0.5246915221214294, + "learning_rate": 2.8499999999999998e-05, + "loss": 0.0499, + "step": 4558 + }, + { + "epoch": 43.0, + "eval_LOCATION_f1": 0.7843137254901962, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.75, + "eval_LOCATION_recall": 0.821917808219178, + "eval_ORGANIZATION_f1": 0.7101449275362318, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6712328767123288, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.8859934853420196, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8662420382165605, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7619047619047619, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7058823529411765, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8571428571428571, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8333333333333334, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.15075072646141052, + "eval_overall_accuracy": 0.9613586020182131, + "eval_overall_f1": 0.8180574555403556, + "eval_overall_precision": 0.7868421052631579, + "eval_overall_recall": 0.8518518518518519, + "eval_runtime": 0.9913, + "eval_samples_per_second": 188.643, + "eval_steps_per_second": 3.026, + "step": 4558 + }, + { + "epoch": 44.0, + "grad_norm": 0.7562664151191711, + "learning_rate": 2.8000000000000003e-05, + "loss": 0.0474, + "step": 4664 + }, + { + "epoch": 44.0, + "eval_LOCATION_f1": 0.8400000000000001, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8181818181818182, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7338129496402878, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6891891891891891, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.887459807073955, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8571428571428571, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7936507936507937, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7352941176470589, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8732394366197184, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8378378378378378, + "eval_TIME_recall": 0.9117647058823529, + "eval_loss": 0.15113013982772827, + "eval_overall_accuracy": 0.9635737140044303, + "eval_overall_f1": 0.8392370572207084, + "eval_overall_precision": 0.804177545691906, + "eval_overall_recall": 0.8774928774928775, + "eval_runtime": 0.9996, + "eval_samples_per_second": 187.066, + "eval_steps_per_second": 3.001, + "step": 4664 + }, + { + "epoch": 45.0, + "grad_norm": 3.280982494354248, + "learning_rate": 2.7500000000000004e-05, + "loss": 0.0499, + "step": 4770 + }, + { + "epoch": 45.0, + "eval_LOCATION_f1": 0.8456375838926175, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8289473684210527, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7375886524822695, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6842105263157895, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.8896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8670886075949367, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7868852459016394, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.75, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8450704225352113, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8108108108108109, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.15040434896945953, + "eval_overall_accuracy": 0.9640659611124784, + "eval_overall_f1": 0.8383561643835615, + "eval_overall_precision": 0.8073878627968337, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.995, + "eval_samples_per_second": 187.933, + "eval_steps_per_second": 3.015, + "step": 4770 + }, + { + "epoch": 46.0, + "grad_norm": 0.43949922919273376, + "learning_rate": 2.7000000000000002e-05, + "loss": 0.0477, + "step": 4876 + }, + { + "epoch": 46.0, + "eval_LOCATION_f1": 0.8133333333333332, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7922077922077922, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.726027397260274, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.654320987654321, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.8896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8670886075949367, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7868852459016394, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.75, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8219178082191781, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7692307692307693, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.15182223916053772, + "eval_overall_accuracy": 0.9625892197883338, + "eval_overall_f1": 0.8265582655826558, + "eval_overall_precision": 0.7881136950904393, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 0.9979, + "eval_samples_per_second": 187.387, + "eval_steps_per_second": 3.006, + "step": 4876 + }, + { + "epoch": 47.0, + "grad_norm": 1.4698158502578735, + "learning_rate": 2.6500000000000004e-05, + "loss": 0.0463, + "step": 4982 + }, + { + "epoch": 47.0, + "eval_LOCATION_f1": 0.8133333333333332, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7922077922077922, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.75, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6551724137931034, + "eval_ORGANIZATION_recall": 0.8769230769230769, + "eval_PERSON_f1": 0.8810289389067525, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8509316770186336, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.75, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6857142857142857, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8055555555555555, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7631578947368421, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.15979216992855072, + "eval_overall_accuracy": 0.9596357371400444, + "eval_overall_f1": 0.822429906542056, + "eval_overall_precision": 0.7738693467336684, + "eval_overall_recall": 0.8774928774928775, + "eval_runtime": 0.9931, + "eval_samples_per_second": 188.295, + "eval_steps_per_second": 3.021, + "step": 4982 + }, + { + "epoch": 48.0, + "grad_norm": 0.8823111057281494, + "learning_rate": 2.6000000000000002e-05, + "loss": 0.0458, + "step": 5088 + }, + { + "epoch": 48.0, + "eval_LOCATION_f1": 0.8076923076923077, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7590361445783133, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7123287671232877, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6419753086419753, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.8867313915857605, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8616352201257862, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8333333333333333, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7894736842105263, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.16684772074222565, + "eval_overall_accuracy": 0.9598818606940684, + "eval_overall_f1": 0.8214765100671142, + "eval_overall_precision": 0.7766497461928934, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.9966, + "eval_samples_per_second": 187.646, + "eval_steps_per_second": 3.01, + "step": 5088 + }, + { + "epoch": 49.0, + "grad_norm": 0.5812783241271973, + "learning_rate": 2.5500000000000003e-05, + "loss": 0.0428, + "step": 5194 + }, + { + "epoch": 49.0, + "eval_LOCATION_f1": 0.8076923076923077, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7590361445783133, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.726027397260274, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.654320987654321, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.8774193548387097, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.85, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8219178082191781, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7692307692307693, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.16307547688484192, + "eval_overall_accuracy": 0.9596357371400444, + "eval_overall_f1": 0.8192771084337349, + "eval_overall_precision": 0.7727272727272727, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.9928, + "eval_samples_per_second": 188.354, + "eval_steps_per_second": 3.022, + "step": 5194 + }, + { + "epoch": 50.0, + "grad_norm": 0.9438253045082092, + "learning_rate": 2.5e-05, + "loss": 0.0437, + "step": 5300 + }, + { + "epoch": 50.0, + "eval_LOCATION_f1": 0.7792207792207793, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7407407407407407, + "eval_LOCATION_recall": 0.821917808219178, + "eval_ORGANIZATION_f1": 0.7272727272727273, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7164179104477612, + "eval_ORGANIZATION_recall": 0.7384615384615385, + "eval_PERSON_f1": 0.8802588996763754, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8553459119496856, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8857142857142858, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8611111111111112, + "eval_TIME_recall": 0.9117647058823529, + "eval_loss": 0.15793509781360626, + "eval_overall_accuracy": 0.9613586020182131, + "eval_overall_f1": 0.8225584594222833, + "eval_overall_precision": 0.7952127659574468, + "eval_overall_recall": 0.8518518518518519, + "eval_runtime": 0.9991, + "eval_samples_per_second": 187.177, + "eval_steps_per_second": 3.003, + "step": 5300 + }, + { + "epoch": 51.0, + "grad_norm": 0.8421459794044495, + "learning_rate": 2.45e-05, + "loss": 0.0436, + "step": 5406 + }, + { + "epoch": 51.0, + "eval_LOCATION_f1": 0.8026315789473685, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7721518987341772, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7092198581560285, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6578947368421053, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8867313915857605, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8616352201257862, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8695652173913043, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8571428571428571, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.15914805233478546, + "eval_overall_accuracy": 0.9613586020182131, + "eval_overall_f1": 0.8240109140518418, + "eval_overall_precision": 0.7905759162303665, + "eval_overall_recall": 0.8603988603988604, + "eval_runtime": 0.9849, + "eval_samples_per_second": 189.872, + "eval_steps_per_second": 3.046, + "step": 5406 + }, + { + "epoch": 52.0, + "grad_norm": 0.6999145150184631, + "learning_rate": 2.4e-05, + "loss": 0.0403, + "step": 5512 + }, + { + "epoch": 52.0, + "eval_LOCATION_f1": 0.8181818181818182, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7777777777777778, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7066666666666667, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6235294117647059, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.8737864077669903, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8490566037735849, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.7619047619047619, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7058823529411765, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8333333333333333, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7894736842105263, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.1684279888868332, + "eval_overall_accuracy": 0.9593896135860202, + "eval_overall_f1": 0.8155080213903743, + "eval_overall_precision": 0.7682619647355163, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 0.9863, + "eval_samples_per_second": 189.596, + "eval_steps_per_second": 3.042, + "step": 5512 + }, + { + "epoch": 53.0, + "grad_norm": 0.8465991616249084, + "learning_rate": 2.35e-05, + "loss": 0.0447, + "step": 5618 + }, + { + "epoch": 53.0, + "eval_LOCATION_f1": 0.8400000000000001, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8181818181818182, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7361111111111112, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6708860759493671, + "eval_ORGANIZATION_recall": 0.8153846153846154, + "eval_PERSON_f1": 0.896774193548387, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.86875, + "eval_PERSON_recall": 0.9266666666666666, + "eval_QUANTITY_f1": 0.7575757575757576, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6756756756756757, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8219178082191781, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7692307692307693, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.16614499688148499, + "eval_overall_accuracy": 0.9606202313561408, + "eval_overall_f1": 0.8344549125168237, + "eval_overall_precision": 0.7908163265306123, + "eval_overall_recall": 0.8831908831908832, + "eval_runtime": 0.9842, + "eval_samples_per_second": 190.011, + "eval_steps_per_second": 3.048, + "step": 5618 + }, + { + "epoch": 54.0, + "grad_norm": 0.7064285278320312, + "learning_rate": 2.3000000000000003e-05, + "loss": 0.0411, + "step": 5724 + }, + { + "epoch": 54.0, + "eval_LOCATION_f1": 0.794701986754967, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7692307692307693, + "eval_LOCATION_recall": 0.821917808219178, + "eval_ORGANIZATION_f1": 0.7183098591549296, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6623376623376623, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8925081433224756, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8726114649681529, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8695652173913043, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8571428571428571, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.16279193758964539, + "eval_overall_accuracy": 0.9603741078021166, + "eval_overall_f1": 0.826265389876881, + "eval_overall_precision": 0.7947368421052632, + "eval_overall_recall": 0.8603988603988604, + "eval_runtime": 0.9899, + "eval_samples_per_second": 188.907, + "eval_steps_per_second": 3.031, + "step": 5724 + }, + { + "epoch": 55.0, + "grad_norm": 0.41567376255989075, + "learning_rate": 2.25e-05, + "loss": 0.039, + "step": 5830 + }, + { + "epoch": 55.0, + "eval_LOCATION_f1": 0.794701986754967, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7692307692307693, + "eval_LOCATION_recall": 0.821917808219178, + "eval_ORGANIZATION_f1": 0.6950354609929077, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6447368421052632, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.8859934853420196, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8662420382165605, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.75, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6857142857142857, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8450704225352113, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8108108108108109, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.16604018211364746, + "eval_overall_accuracy": 0.9581589958158996, + "eval_overall_f1": 0.8147138964577657, + "eval_overall_precision": 0.7806788511749347, + "eval_overall_recall": 0.8518518518518519, + "eval_runtime": 0.9924, + "eval_samples_per_second": 188.441, + "eval_steps_per_second": 3.023, + "step": 5830 + }, + { + "epoch": 56.0, + "grad_norm": 3.736433744430542, + "learning_rate": 2.2000000000000003e-05, + "loss": 0.0416, + "step": 5936 + }, + { + "epoch": 56.0, + "eval_LOCATION_f1": 0.8157894736842105, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7848101265822784, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7246376811594203, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.684931506849315, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8932038834951457, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8679245283018868, + "eval_PERSON_recall": 0.92, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8695652173913043, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8571428571428571, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.1560632586479187, + "eval_overall_accuracy": 0.9628353433423579, + "eval_overall_f1": 0.832876712328767, + "eval_overall_precision": 0.8021108179419525, + "eval_overall_recall": 0.8660968660968661, + "eval_runtime": 0.9995, + "eval_samples_per_second": 187.088, + "eval_steps_per_second": 3.001, + "step": 5936 + }, + { + "epoch": 57.0, + "grad_norm": 0.16218724846839905, + "learning_rate": 2.15e-05, + "loss": 0.0386, + "step": 6042 + }, + { + "epoch": 57.0, + "eval_LOCATION_f1": 0.8051948051948052, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7654320987654321, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7313432835820897, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7101449275362319, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.8802588996763754, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8553459119496856, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7812500000000001, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7142857142857143, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8333333333333333, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7894736842105263, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.1614658534526825, + "eval_overall_accuracy": 0.9616047255722373, + "eval_overall_f1": 0.8240109140518418, + "eval_overall_precision": 0.7905759162303665, + "eval_overall_recall": 0.8603988603988604, + "eval_runtime": 0.9919, + "eval_samples_per_second": 188.532, + "eval_steps_per_second": 3.025, + "step": 6042 + }, + { + "epoch": 58.0, + "grad_norm": 0.9404721856117249, + "learning_rate": 2.1e-05, + "loss": 0.0395, + "step": 6148 + }, + { + "epoch": 58.0, + "eval_LOCATION_f1": 0.8076923076923077, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7590361445783133, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7313432835820897, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7101449275362319, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.8867313915857605, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8616352201257862, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7619047619047619, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7058823529411765, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8695652173913043, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8571428571428571, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.1566106230020523, + "eval_overall_accuracy": 0.9625892197883338, + "eval_overall_f1": 0.8290013679890562, + "eval_overall_precision": 0.7973684210526316, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 0.9978, + "eval_samples_per_second": 187.413, + "eval_steps_per_second": 3.007, + "step": 6148 + }, + { + "epoch": 59.0, + "grad_norm": 1.2607734203338623, + "learning_rate": 2.05e-05, + "loss": 0.0376, + "step": 6254 + }, + { + "epoch": 59.0, + "eval_LOCATION_f1": 0.8258064516129032, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7804878048780488, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7218045112781954, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7058823529411765, + "eval_ORGANIZATION_recall": 0.7384615384615385, + "eval_PERSON_f1": 0.8701298701298702, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8481012658227848, + "eval_PERSON_recall": 0.8933333333333333, + "eval_QUANTITY_f1": 0.7868852459016394, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.75, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8571428571428571, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8333333333333334, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.16003942489624023, + "eval_overall_accuracy": 0.9633275904504061, + "eval_overall_f1": 0.8253094910591472, + "eval_overall_precision": 0.7978723404255319, + "eval_overall_recall": 0.8547008547008547, + "eval_runtime": 1.001, + "eval_samples_per_second": 186.811, + "eval_steps_per_second": 2.997, + "step": 6254 + }, + { + "epoch": 60.0, + "grad_norm": 0.5839110612869263, + "learning_rate": 2e-05, + "loss": 0.0364, + "step": 6360 + }, + { + "epoch": 60.0, + "eval_LOCATION_f1": 0.8378378378378378, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8266666666666667, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7205882352941175, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6901408450704225, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.8758169934640523, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8589743589743589, + "eval_PERSON_recall": 0.8933333333333333, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8695652173913043, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8571428571428571, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.15972208976745605, + "eval_overall_accuracy": 0.9633275904504061, + "eval_overall_f1": 0.8294036061026352, + "eval_overall_precision": 0.8081081081081081, + "eval_overall_recall": 0.8518518518518519, + "eval_runtime": 0.9927, + "eval_samples_per_second": 188.383, + "eval_steps_per_second": 3.022, + "step": 6360 + }, + { + "epoch": 61.0, + "grad_norm": 0.9700207710266113, + "learning_rate": 1.9500000000000003e-05, + "loss": 0.0362, + "step": 6466 + }, + { + "epoch": 61.0, + "eval_LOCATION_f1": 0.8513513513513513, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.84, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7132867132867133, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6538461538461539, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8670886075949367, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8571428571428571, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8333333333333334, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.16418209671974182, + "eval_overall_accuracy": 0.9618508491262614, + "eval_overall_f1": 0.8344733242134063, + "eval_overall_precision": 0.8026315789473685, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 0.9901, + "eval_samples_per_second": 188.877, + "eval_steps_per_second": 3.03, + "step": 6466 + }, + { + "epoch": 62.0, + "grad_norm": 0.7778552770614624, + "learning_rate": 1.9e-05, + "loss": 0.0358, + "step": 6572 + }, + { + "epoch": 62.0, + "eval_LOCATION_f1": 0.8366013071895425, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7338129496402878, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6891891891891891, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8888888888888887, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8717948717948718, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7692307692307692, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.6944444444444444, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8571428571428571, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8333333333333334, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.1708838939666748, + "eval_overall_accuracy": 0.9613586020182131, + "eval_overall_f1": 0.834924965893588, + "eval_overall_precision": 0.8010471204188482, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 1.08, + "eval_samples_per_second": 173.149, + "eval_steps_per_second": 2.778, + "step": 6572 + }, + { + "epoch": 63.0, + "grad_norm": 2.994206428527832, + "learning_rate": 1.85e-05, + "loss": 0.034, + "step": 6678 + }, + { + "epoch": 63.0, + "eval_LOCATION_f1": 0.8571428571428572, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8513513513513513, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7338129496402878, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6891891891891891, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8729641693811074, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8535031847133758, + "eval_PERSON_recall": 0.8933333333333333, + "eval_QUANTITY_f1": 0.8196721311475409, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.78125, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8695652173913043, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8571428571428571, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.164272740483284, + "eval_overall_accuracy": 0.9635737140044303, + "eval_overall_f1": 0.8381742738589213, + "eval_overall_precision": 0.8145161290322581, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 0.9945, + "eval_samples_per_second": 188.026, + "eval_steps_per_second": 3.016, + "step": 6678 + }, + { + "epoch": 64.0, + "grad_norm": 0.950914740562439, + "learning_rate": 1.8e-05, + "loss": 0.0353, + "step": 6784 + }, + { + "epoch": 64.0, + "eval_LOCATION_f1": 0.8441558441558442, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8024691358024691, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7272727272727273, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7164179104477612, + "eval_ORGANIZATION_recall": 0.7384615384615385, + "eval_PERSON_f1": 0.8758169934640523, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8589743589743589, + "eval_PERSON_recall": 0.8933333333333333, + "eval_QUANTITY_f1": 0.7868852459016394, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.75, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8695652173913043, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8571428571428571, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.16246764361858368, + "eval_overall_accuracy": 0.9635737140044303, + "eval_overall_f1": 0.8337950138504155, + "eval_overall_precision": 0.8113207547169812, + "eval_overall_recall": 0.8575498575498576, + "eval_runtime": 0.994, + "eval_samples_per_second": 188.133, + "eval_steps_per_second": 3.018, + "step": 6784 + }, + { + "epoch": 65.0, + "grad_norm": 0.5340273976325989, + "learning_rate": 1.75e-05, + "loss": 0.0335, + "step": 6890 + }, + { + "epoch": 65.0, + "eval_LOCATION_f1": 0.8441558441558442, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8024691358024691, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7407407407407408, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7142857142857143, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8766233766233766, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8544303797468354, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7317073170731707, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.17117133736610413, + "eval_overall_accuracy": 0.9628353433423579, + "eval_overall_f1": 0.8283378746594006, + "eval_overall_precision": 0.793733681462141, + "eval_overall_recall": 0.8660968660968661, + "eval_runtime": 0.9791, + "eval_samples_per_second": 190.991, + "eval_steps_per_second": 3.064, + "step": 6890 + }, + { + "epoch": 66.0, + "grad_norm": 0.4983876049518585, + "learning_rate": 1.7000000000000003e-05, + "loss": 0.0344, + "step": 6996 + }, + { + "epoch": 66.0, + "eval_LOCATION_f1": 0.8344370860927152, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8076923076923077, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7428571428571429, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6933333333333334, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.8838709677419355, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.85625, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8333333333333333, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.7894736842105263, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.16945300996303558, + "eval_overall_accuracy": 0.9623430962343096, + "eval_overall_f1": 0.8326530612244898, + "eval_overall_precision": 0.796875, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.9853, + "eval_samples_per_second": 189.785, + "eval_steps_per_second": 3.045, + "step": 6996 + }, + { + "epoch": 67.0, + "grad_norm": 0.8631144762039185, + "learning_rate": 1.65e-05, + "loss": 0.0333, + "step": 7102 + }, + { + "epoch": 67.0, + "eval_LOCATION_f1": 0.8441558441558442, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8024691358024691, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7246376811594203, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.684931506849315, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8766233766233766, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8544303797468354, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.8064516129032258, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7575757575757576, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1660642921924591, + "eval_overall_accuracy": 0.9633275904504061, + "eval_overall_f1": 0.8317373461012312, + "eval_overall_precision": 0.8, + "eval_overall_recall": 0.8660968660968661, + "eval_runtime": 0.9933, + "eval_samples_per_second": 188.27, + "eval_steps_per_second": 3.02, + "step": 7102 + }, + { + "epoch": 68.0, + "grad_norm": 0.16754050552845, + "learning_rate": 1.6000000000000003e-05, + "loss": 0.0344, + "step": 7208 + }, + { + "epoch": 68.0, + "eval_LOCATION_f1": 0.8441558441558442, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8024691358024691, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7313432835820897, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7101449275362319, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.8794788273615636, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8598726114649682, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.8064516129032258, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7575757575757576, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1679677516222, + "eval_overall_accuracy": 0.9638198375584544, + "eval_overall_f1": 0.8347107438016529, + "eval_overall_precision": 0.808, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 0.9969, + "eval_samples_per_second": 187.58, + "eval_steps_per_second": 3.009, + "step": 7208 + }, + { + "epoch": 69.0, + "grad_norm": 0.08073262125253677, + "learning_rate": 1.55e-05, + "loss": 0.0297, + "step": 7314 + }, + { + "epoch": 69.0, + "eval_LOCATION_f1": 0.8366013071895425, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7222222222222222, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6582278481012658, + "eval_ORGANIZATION_recall": 0.8, + "eval_PERSON_f1": 0.8802588996763754, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8553459119496856, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7936507936507937, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7352941176470589, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17325514554977417, + "eval_overall_accuracy": 0.9618508491262614, + "eval_overall_f1": 0.8292682926829267, + "eval_overall_precision": 0.7906976744186046, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.9946, + "eval_samples_per_second": 188.01, + "eval_steps_per_second": 3.016, + "step": 7314 + }, + { + "epoch": 70.0, + "grad_norm": 0.45353350043296814, + "learning_rate": 1.5e-05, + "loss": 0.0304, + "step": 7420 + }, + { + "epoch": 70.0, + "eval_LOCATION_f1": 0.8421052631578948, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.810126582278481, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.75, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7183098591549296, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8729641693811074, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8535031847133758, + "eval_PERSON_recall": 0.8933333333333333, + "eval_QUANTITY_f1": 0.8196721311475409, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.78125, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17020219564437866, + "eval_overall_accuracy": 0.9638198375584544, + "eval_overall_f1": 0.8358620689655173, + "eval_overall_precision": 0.8101604278074866, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 0.9973, + "eval_samples_per_second": 187.51, + "eval_steps_per_second": 3.008, + "step": 7420 + }, + { + "epoch": 71.0, + "grad_norm": 1.6932181119918823, + "learning_rate": 1.45e-05, + "loss": 0.0329, + "step": 7526 + }, + { + "epoch": 71.0, + "eval_LOCATION_f1": 0.8333333333333334, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7831325301204819, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7050359712230215, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6621621621621622, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.8701298701298702, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8481012658227848, + "eval_PERSON_recall": 0.8933333333333333, + "eval_QUANTITY_f1": 0.8196721311475409, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.78125, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17471523582935333, + "eval_overall_accuracy": 0.963081466896382, + "eval_overall_f1": 0.8240109140518418, + "eval_overall_precision": 0.7905759162303665, + "eval_overall_recall": 0.8603988603988604, + "eval_runtime": 0.9916, + "eval_samples_per_second": 188.583, + "eval_steps_per_second": 3.025, + "step": 7526 + }, + { + "epoch": 72.0, + "grad_norm": 0.46643197536468506, + "learning_rate": 1.4000000000000001e-05, + "loss": 0.0319, + "step": 7632 + }, + { + "epoch": 72.0, + "eval_LOCATION_f1": 0.8299319727891157, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8243243243243243, + "eval_LOCATION_recall": 0.8356164383561644, + "eval_ORGANIZATION_f1": 0.7194244604316546, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6756756756756757, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8673139158576051, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8427672955974843, + "eval_PERSON_recall": 0.8933333333333333, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.16767063736915588, + "eval_overall_accuracy": 0.9618508491262614, + "eval_overall_f1": 0.8209366391184574, + "eval_overall_precision": 0.7946666666666666, + "eval_overall_recall": 0.8490028490028491, + "eval_runtime": 1.0071, + "eval_samples_per_second": 185.686, + "eval_steps_per_second": 2.979, + "step": 7632 + }, + { + "epoch": 73.0, + "grad_norm": 0.1150897815823555, + "learning_rate": 1.3500000000000001e-05, + "loss": 0.0321, + "step": 7738 + }, + { + "epoch": 73.0, + "eval_LOCATION_f1": 0.8513513513513513, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.84, + "eval_LOCATION_recall": 0.863013698630137, + "eval_ORGANIZATION_f1": 0.7194244604316546, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6756756756756757, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8766233766233766, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8544303797468354, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.8196721311475409, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.78125, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8695652173913043, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8571428571428571, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.17036782205104828, + "eval_overall_accuracy": 0.9645582082205267, + "eval_overall_f1": 0.8358620689655173, + "eval_overall_precision": 0.8101604278074866, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 1.0038, + "eval_samples_per_second": 186.293, + "eval_steps_per_second": 2.989, + "step": 7738 + }, + { + "epoch": 74.0, + "grad_norm": 0.2557017505168915, + "learning_rate": 1.3000000000000001e-05, + "loss": 0.0306, + "step": 7844 + }, + { + "epoch": 74.0, + "eval_LOCATION_f1": 0.8435374149659863, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8378378378378378, + "eval_LOCATION_recall": 0.8493150684931506, + "eval_ORGANIZATION_f1": 0.7101449275362318, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6712328767123288, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.8745980707395499, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.84472049689441, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7812500000000001, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7142857142857143, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8571428571428571, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8333333333333334, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.1716192364692688, + "eval_overall_accuracy": 0.9628353433423579, + "eval_overall_f1": 0.8273972602739725, + "eval_overall_precision": 0.7968337730870713, + "eval_overall_recall": 0.8603988603988604, + "eval_runtime": 0.9992, + "eval_samples_per_second": 187.155, + "eval_steps_per_second": 3.002, + "step": 7844 + }, + { + "epoch": 75.0, + "grad_norm": 0.24300232529640198, + "learning_rate": 1.25e-05, + "loss": 0.0299, + "step": 7950 + }, + { + "epoch": 75.0, + "eval_LOCATION_f1": 0.8387096774193549, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7926829268292683, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7338129496402878, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6891891891891891, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8802588996763754, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8553459119496856, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.8387096774193549, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7878787878787878, + "eval_QUANTITY_recall": 0.896551724137931, + "eval_TIME_f1": 0.8695652173913043, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8571428571428571, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.17053084075450897, + "eval_overall_accuracy": 0.9640659611124784, + "eval_overall_f1": 0.8392370572207084, + "eval_overall_precision": 0.804177545691906, + "eval_overall_recall": 0.8774928774928775, + "eval_runtime": 0.9826, + "eval_samples_per_second": 190.304, + "eval_steps_per_second": 3.053, + "step": 7950 + }, + { + "epoch": 76.0, + "grad_norm": 0.3287041485309601, + "learning_rate": 1.2e-05, + "loss": 0.0283, + "step": 8056 + }, + { + "epoch": 76.0, + "eval_LOCATION_f1": 0.8387096774193549, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7926829268292683, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7234042553191491, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6710526315789473, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8831168831168831, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8607594936708861, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8571428571428571, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8333333333333334, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.1747513860464096, + "eval_overall_accuracy": 0.9620969726802855, + "eval_overall_f1": 0.8315217391304348, + "eval_overall_precision": 0.7948051948051948, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.9912, + "eval_samples_per_second": 188.663, + "eval_steps_per_second": 3.027, + "step": 8056 + }, + { + "epoch": 77.0, + "grad_norm": 1.2225937843322754, + "learning_rate": 1.1500000000000002e-05, + "loss": 0.0303, + "step": 8162 + }, + { + "epoch": 77.0, + "eval_LOCATION_f1": 0.8333333333333334, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7831325301204819, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.72992700729927, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6944444444444444, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8802588996763754, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8553459119496856, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7868852459016394, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.75, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8695652173913043, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8571428571428571, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.17066791653633118, + "eval_overall_accuracy": 0.963081466896382, + "eval_overall_f1": 0.8333333333333334, + "eval_overall_precision": 0.800524934383202, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 0.9877, + "eval_samples_per_second": 189.325, + "eval_steps_per_second": 3.037, + "step": 8162 + }, + { + "epoch": 78.0, + "grad_norm": 1.3090835809707642, + "learning_rate": 1.1000000000000001e-05, + "loss": 0.0279, + "step": 8268 + }, + { + "epoch": 78.0, + "eval_LOCATION_f1": 0.8387096774193549, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7926829268292683, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7205882352941175, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6901408450704225, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.8896103896103896, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8670886075949367, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.8064516129032258, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7575757575757576, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17071382701396942, + "eval_overall_accuracy": 0.9635737140044303, + "eval_overall_f1": 0.8356164383561645, + "eval_overall_precision": 0.8047493403693932, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 0.9773, + "eval_samples_per_second": 191.338, + "eval_steps_per_second": 3.07, + "step": 8268 + }, + { + "epoch": 79.0, + "grad_norm": 0.48243507742881775, + "learning_rate": 1.05e-05, + "loss": 0.0291, + "step": 8374 + }, + { + "epoch": 79.0, + "eval_LOCATION_f1": 0.8496732026143791, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8125, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7313432835820897, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7101449275362319, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.8802588996763754, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8553459119496856, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17151491343975067, + "eval_overall_accuracy": 0.9643120846665025, + "eval_overall_f1": 0.8324175824175825, + "eval_overall_precision": 0.8037135278514589, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 0.9871, + "eval_samples_per_second": 189.447, + "eval_steps_per_second": 3.039, + "step": 8374 + }, + { + "epoch": 80.0, + "grad_norm": 0.3613625466823578, + "learning_rate": 1e-05, + "loss": 0.0291, + "step": 8480 + }, + { + "epoch": 80.0, + "eval_LOCATION_f1": 0.846153846153846, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7951807228915663, + "eval_LOCATION_recall": 0.9041095890410958, + "eval_ORGANIZATION_f1": 0.7194244604316546, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6756756756756757, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8823529411764707, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.7868852459016394, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.75, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8695652173913043, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8571428571428571, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.17464391887187958, + "eval_overall_accuracy": 0.963081466896382, + "eval_overall_f1": 0.8344733242134063, + "eval_overall_precision": 0.8026315789473685, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 1.0025, + "eval_samples_per_second": 186.532, + "eval_steps_per_second": 2.992, + "step": 8480 + }, + { + "epoch": 81.0, + "grad_norm": 0.47663748264312744, + "learning_rate": 9.5e-06, + "loss": 0.0264, + "step": 8586 + }, + { + "epoch": 81.0, + "eval_LOCATION_f1": 0.8333333333333334, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7831325301204819, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7205882352941175, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6901408450704225, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.8823529411764707, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17564432322978973, + "eval_overall_accuracy": 0.9635737140044303, + "eval_overall_f1": 0.8285322359396434, + "eval_overall_precision": 0.798941798941799, + "eval_overall_recall": 0.8603988603988604, + "eval_runtime": 0.9953, + "eval_samples_per_second": 187.89, + "eval_steps_per_second": 3.014, + "step": 8586 + }, + { + "epoch": 82.0, + "grad_norm": 0.739096462726593, + "learning_rate": 9e-06, + "loss": 0.0266, + "step": 8692 + }, + { + "epoch": 82.0, + "eval_LOCATION_f1": 0.8333333333333334, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7831325301204819, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.725925925925926, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.8823529411764707, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8653846153846154, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.7868852459016394, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.75, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17531628906726837, + "eval_overall_accuracy": 0.9643120846665025, + "eval_overall_f1": 0.8308115543328748, + "eval_overall_precision": 0.8031914893617021, + "eval_overall_recall": 0.8603988603988604, + "eval_runtime": 0.9906, + "eval_samples_per_second": 188.775, + "eval_steps_per_second": 3.028, + "step": 8692 + }, + { + "epoch": 83.0, + "grad_norm": 3.826598644256592, + "learning_rate": 8.500000000000002e-06, + "loss": 0.0248, + "step": 8798 + }, + { + "epoch": 83.0, + "eval_LOCATION_f1": 0.8441558441558442, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8024691358024691, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7480916030534351, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7424242424242424, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.8729641693811074, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8535031847133758, + "eval_PERSON_recall": 0.8933333333333333, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17535383999347687, + "eval_overall_accuracy": 0.9645582082205267, + "eval_overall_f1": 0.8326417704011064, + "eval_overall_precision": 0.8091397849462365, + "eval_overall_recall": 0.8575498575498576, + "eval_runtime": 1.0064, + "eval_samples_per_second": 185.819, + "eval_steps_per_second": 2.981, + "step": 8798 + }, + { + "epoch": 84.0, + "grad_norm": 0.8354291915893555, + "learning_rate": 8.000000000000001e-06, + "loss": 0.027, + "step": 8904 + }, + { + "epoch": 84.0, + "eval_LOCATION_f1": 0.8609271523178808, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8333333333333334, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7246376811594203, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.684931506849315, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8831168831168831, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8607594936708861, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7868852459016394, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.75, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8695652173913043, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8571428571428571, + "eval_TIME_recall": 0.8823529411764706, + "eval_loss": 0.1745665818452835, + "eval_overall_accuracy": 0.9648043317745508, + "eval_overall_f1": 0.8390646492434664, + "eval_overall_precision": 0.8111702127659575, + "eval_overall_recall": 0.8689458689458689, + "eval_runtime": 1.0704, + "eval_samples_per_second": 174.7, + "eval_steps_per_second": 2.803, + "step": 8904 + }, + { + "epoch": 85.0, + "grad_norm": 1.275327444076538, + "learning_rate": 7.5e-06, + "loss": 0.0278, + "step": 9010 + }, + { + "epoch": 85.0, + "eval_LOCATION_f1": 0.8387096774193549, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7926829268292683, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7153284671532848, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6805555555555556, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.8810289389067525, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8509316770186336, + "eval_PERSON_recall": 0.9133333333333333, + "eval_QUANTITY_f1": 0.7868852459016394, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.75, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.18171997368335724, + "eval_overall_accuracy": 0.9628353433423579, + "eval_overall_f1": 0.8283378746594006, + "eval_overall_precision": 0.793733681462141, + "eval_overall_recall": 0.8660968660968661, + "eval_runtime": 0.988, + "eval_samples_per_second": 189.269, + "eval_steps_per_second": 3.036, + "step": 9010 + }, + { + "epoch": 86.0, + "grad_norm": 0.6492407321929932, + "learning_rate": 7.000000000000001e-06, + "loss": 0.0266, + "step": 9116 + }, + { + "epoch": 86.0, + "eval_LOCATION_f1": 0.8387096774193549, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7926829268292683, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7313432835820897, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7101449275362319, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.8766233766233766, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8544303797468354, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.8196721311475409, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.78125, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1775309443473816, + "eval_overall_accuracy": 0.9648043317745508, + "eval_overall_f1": 0.8324175824175825, + "eval_overall_precision": 0.8037135278514589, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 0.9952, + "eval_samples_per_second": 187.906, + "eval_steps_per_second": 3.015, + "step": 9116 + }, + { + "epoch": 87.0, + "grad_norm": 0.48030099272727966, + "learning_rate": 6.5000000000000004e-06, + "loss": 0.0261, + "step": 9222 + }, + { + "epoch": 87.0, + "eval_LOCATION_f1": 0.8533333333333334, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8311688311688312, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7183098591549296, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6623376623376623, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8859934853420196, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8662420382165605, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7868852459016394, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.75, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17771387100219727, + "eval_overall_accuracy": 0.9640659611124784, + "eval_overall_f1": 0.8340192043895748, + "eval_overall_precision": 0.8042328042328042, + "eval_overall_recall": 0.8660968660968661, + "eval_runtime": 0.9953, + "eval_samples_per_second": 187.876, + "eval_steps_per_second": 3.014, + "step": 9222 + }, + { + "epoch": 88.0, + "grad_norm": 0.7421311736106873, + "learning_rate": 6e-06, + "loss": 0.0276, + "step": 9328 + }, + { + "epoch": 88.0, + "eval_LOCATION_f1": 0.847682119205298, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8205128205128205, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.72992700729927, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6944444444444444, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8794788273615636, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8598726114649682, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1769006848335266, + "eval_overall_accuracy": 0.9633275904504061, + "eval_overall_f1": 0.8319559228650139, + "eval_overall_precision": 0.8053333333333333, + "eval_overall_recall": 0.8603988603988604, + "eval_runtime": 0.998, + "eval_samples_per_second": 187.379, + "eval_steps_per_second": 3.006, + "step": 9328 + }, + { + "epoch": 89.0, + "grad_norm": 0.4191894829273224, + "learning_rate": 5.500000000000001e-06, + "loss": 0.0261, + "step": 9434 + }, + { + "epoch": 89.0, + "eval_LOCATION_f1": 0.8258064516129032, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7804878048780488, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7153284671532848, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6805555555555556, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.8794788273615636, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8598726114649682, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.7868852459016394, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.75, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1796666532754898, + "eval_overall_accuracy": 0.963081466896382, + "eval_overall_f1": 0.8257887517146777, + "eval_overall_precision": 0.7962962962962963, + "eval_overall_recall": 0.8575498575498576, + "eval_runtime": 0.9969, + "eval_samples_per_second": 187.575, + "eval_steps_per_second": 3.009, + "step": 9434 + }, + { + "epoch": 90.0, + "grad_norm": 0.639091968536377, + "learning_rate": 5e-06, + "loss": 0.0251, + "step": 9540 + }, + { + "epoch": 90.0, + "eval_LOCATION_f1": 0.8496732026143791, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8125, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7352941176470589, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.704225352112676, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8859934853420196, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8662420382165605, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7868852459016394, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.75, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17721274495124817, + "eval_overall_accuracy": 0.9650504553285749, + "eval_overall_f1": 0.837465564738292, + "eval_overall_precision": 0.8106666666666666, + "eval_overall_recall": 0.8660968660968661, + "eval_runtime": 0.9991, + "eval_samples_per_second": 187.173, + "eval_steps_per_second": 3.003, + "step": 9540 + }, + { + "epoch": 91.0, + "grad_norm": 0.6990874409675598, + "learning_rate": 4.5e-06, + "loss": 0.0243, + "step": 9646 + }, + { + "epoch": 91.0, + "eval_LOCATION_f1": 0.8311688311688312, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.7901234567901234, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7368421052631577, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7205882352941176, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.8794788273615636, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8598726114649682, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.7741935483870968, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.7272727272727273, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17922192811965942, + "eval_overall_accuracy": 0.9645582082205267, + "eval_overall_f1": 0.830344827586207, + "eval_overall_precision": 0.8048128342245989, + "eval_overall_recall": 0.8575498575498576, + "eval_runtime": 0.9978, + "eval_samples_per_second": 187.404, + "eval_steps_per_second": 3.006, + "step": 9646 + }, + { + "epoch": 92.0, + "grad_norm": 0.31417253613471985, + "learning_rate": 4.000000000000001e-06, + "loss": 0.0246, + "step": 9752 + }, + { + "epoch": 92.0, + "eval_LOCATION_f1": 0.8496732026143791, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8125, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.725925925925926, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.7, + "eval_ORGANIZATION_recall": 0.7538461538461538, + "eval_PERSON_f1": 0.8794788273615636, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8598726114649682, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.7868852459016394, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.75, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17927689850330353, + "eval_overall_accuracy": 0.9635737140044303, + "eval_overall_f1": 0.833103448275862, + "eval_overall_precision": 0.8074866310160428, + "eval_overall_recall": 0.8603988603988604, + "eval_runtime": 1.0031, + "eval_samples_per_second": 186.424, + "eval_steps_per_second": 2.991, + "step": 9752 + }, + { + "epoch": 93.0, + "grad_norm": 0.2049255222082138, + "learning_rate": 3.5000000000000004e-06, + "loss": 0.0252, + "step": 9858 + }, + { + "epoch": 93.0, + "eval_LOCATION_f1": 0.8609271523178808, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8333333333333334, + "eval_LOCATION_recall": 0.8904109589041096, + "eval_ORGANIZATION_f1": 0.7338129496402878, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6891891891891891, + "eval_ORGANIZATION_recall": 0.7846153846153846, + "eval_PERSON_f1": 0.8859934853420196, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8662420382165605, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.8196721311475409, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.78125, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1780104637145996, + "eval_overall_accuracy": 0.9640659611124784, + "eval_overall_f1": 0.8418156808803302, + "eval_overall_precision": 0.8138297872340425, + "eval_overall_recall": 0.8717948717948718, + "eval_runtime": 0.9912, + "eval_samples_per_second": 188.665, + "eval_steps_per_second": 3.027, + "step": 9858 + }, + { + "epoch": 94.0, + "grad_norm": 0.3145851790904999, + "learning_rate": 3e-06, + "loss": 0.0252, + "step": 9964 + }, + { + "epoch": 94.0, + "eval_LOCATION_f1": 0.8533333333333334, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8311688311688312, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7352941176470589, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.704225352112676, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8794788273615636, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8598726114649682, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.8196721311475409, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.78125, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17691487073898315, + "eval_overall_accuracy": 0.9640659611124784, + "eval_overall_f1": 0.8381742738589213, + "eval_overall_precision": 0.8145161290322581, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 0.9959, + "eval_samples_per_second": 187.776, + "eval_steps_per_second": 3.012, + "step": 9964 + }, + { + "epoch": 95.0, + "grad_norm": 0.3581851124763489, + "learning_rate": 2.5e-06, + "loss": 0.0241, + "step": 10070 + }, + { + "epoch": 95.0, + "eval_LOCATION_f1": 0.8533333333333334, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8311688311688312, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.72992700729927, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.6944444444444444, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8859934853420196, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8662420382165605, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.8196721311475409, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.78125, + "eval_QUANTITY_recall": 0.8620689655172413, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17827032506465912, + "eval_overall_accuracy": 0.9635737140044303, + "eval_overall_f1": 0.8397790055248618, + "eval_overall_precision": 0.8150134048257373, + "eval_overall_recall": 0.8660968660968661, + "eval_runtime": 1.0048, + "eval_samples_per_second": 186.108, + "eval_steps_per_second": 2.986, + "step": 10070 + }, + { + "epoch": 96.0, + "grad_norm": 0.5018787384033203, + "learning_rate": 2.0000000000000003e-06, + "loss": 0.0239, + "step": 10176 + }, + { + "epoch": 96.0, + "eval_LOCATION_f1": 0.8533333333333334, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8311688311688312, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7352941176470589, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.704225352112676, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8766233766233766, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8544303797468354, + "eval_PERSON_recall": 0.9, + "eval_QUANTITY_f1": 0.7868852459016394, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.75, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1762346476316452, + "eval_overall_accuracy": 0.9638198375584544, + "eval_overall_f1": 0.8342541436464089, + "eval_overall_precision": 0.8096514745308311, + "eval_overall_recall": 0.8603988603988604, + "eval_runtime": 0.9962, + "eval_samples_per_second": 187.711, + "eval_steps_per_second": 3.011, + "step": 10176 + }, + { + "epoch": 97.0, + "grad_norm": 0.7310182452201843, + "learning_rate": 1.5e-06, + "loss": 0.0226, + "step": 10282 + }, + { + "epoch": 97.0, + "eval_LOCATION_f1": 0.8533333333333334, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8311688311688312, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7352941176470589, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.704225352112676, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8859934853420196, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8662420382165605, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7868852459016394, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.75, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17845137417316437, + "eval_overall_accuracy": 0.9643120846665025, + "eval_overall_f1": 0.8381742738589213, + "eval_overall_precision": 0.8145161290322581, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 1.0543, + "eval_samples_per_second": 177.368, + "eval_steps_per_second": 2.845, + "step": 10282 + }, + { + "epoch": 98.0, + "grad_norm": 0.48724424839019775, + "learning_rate": 1.0000000000000002e-06, + "loss": 0.0244, + "step": 10388 + }, + { + "epoch": 98.0, + "eval_LOCATION_f1": 0.8533333333333334, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8311688311688312, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7352941176470589, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.704225352112676, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8859934853420196, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8662420382165605, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7868852459016394, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.75, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8285714285714286, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8055555555555556, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17859233915805817, + "eval_overall_accuracy": 0.9640659611124784, + "eval_overall_f1": 0.8370165745856354, + "eval_overall_precision": 0.8123324396782842, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 0.9905, + "eval_samples_per_second": 188.793, + "eval_steps_per_second": 3.029, + "step": 10388 + }, + { + "epoch": 99.0, + "grad_norm": 0.3531472086906433, + "learning_rate": 5.000000000000001e-07, + "loss": 0.0244, + "step": 10494 + }, + { + "epoch": 99.0, + "eval_LOCATION_f1": 0.8533333333333334, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8311688311688312, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7352941176470589, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.704225352112676, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8859934853420196, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8662420382165605, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7868852459016394, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.75, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.17733581364154816, + "eval_overall_accuracy": 0.9645582082205267, + "eval_overall_f1": 0.8381742738589213, + "eval_overall_precision": 0.8145161290322581, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 0.992, + "eval_samples_per_second": 188.514, + "eval_steps_per_second": 3.024, + "step": 10494 + }, + { + "epoch": 100.0, + "grad_norm": 0.8635873794555664, + "learning_rate": 0.0, + "loss": 0.0265, + "step": 10600 + }, + { + "epoch": 100.0, + "eval_LOCATION_f1": 0.8533333333333334, + "eval_LOCATION_number": 73, + "eval_LOCATION_precision": 0.8311688311688312, + "eval_LOCATION_recall": 0.8767123287671232, + "eval_ORGANIZATION_f1": 0.7352941176470589, + "eval_ORGANIZATION_number": 65, + "eval_ORGANIZATION_precision": 0.704225352112676, + "eval_ORGANIZATION_recall": 0.7692307692307693, + "eval_PERSON_f1": 0.8859934853420196, + "eval_PERSON_number": 150, + "eval_PERSON_precision": 0.8662420382165605, + "eval_PERSON_recall": 0.9066666666666666, + "eval_QUANTITY_f1": 0.7868852459016394, + "eval_QUANTITY_number": 29, + "eval_QUANTITY_precision": 0.75, + "eval_QUANTITY_recall": 0.8275862068965517, + "eval_TIME_f1": 0.8405797101449276, + "eval_TIME_number": 34, + "eval_TIME_precision": 0.8285714285714286, + "eval_TIME_recall": 0.8529411764705882, + "eval_loss": 0.1773531436920166, + "eval_overall_accuracy": 0.9643120846665025, + "eval_overall_f1": 0.8381742738589213, + "eval_overall_precision": 0.8145161290322581, + "eval_overall_recall": 0.8632478632478633, + "eval_runtime": 0.9958, + "eval_samples_per_second": 187.787, + "eval_steps_per_second": 3.013, + "step": 10600 + }, + { + "epoch": 100.0, + "step": 10600, + "total_flos": 5043631411410816.0, + "train_loss": 0.07336943858074692, + "train_runtime": 1365.3768, + "train_samples_per_second": 123.629, + "train_steps_per_second": 7.763 + } + ], + "logging_steps": 500, + "max_steps": 10600, + "num_input_tokens_seen": 0, + "num_train_epochs": 100, + "save_steps": 500, + "total_flos": 5043631411410816.0, + "train_batch_size": 16, + "trial_name": null, + "trial_params": null +}